Jeff Sonas has just done some work comparing Carlsen's recent great performance in Nanjing to other hisorically superlative performances.

Interesting reading.

http://www.chessbase.com/newsdetail.asp?newsid=5828

Best Tournament Performances since January 2005# Player Score/% opp. Perf. Event

1
Magnus Carlsen
 8/10 (80%) 2733 2850
Nanjing, 2009

2
Vassily Ivanchuk
 8/10 (80%) 2715 2835
Mtel Masters, Sofia, 2008

3
Veselin Topalov
10/14 (71%) 2726 2830
San Luis (World Championship), 2005

4
Veselin Topalov
 8/12 (67%) 2741 2804
Linares, 2005

5
Garry Kasparov
 8/12 (67%) 2733 2797
Linares, 2005

6
Viswanathan Anand
 9/13 (69%) 2706 2794
Corus A, 2006

 
Veselin Topalov
 9/13 (69%) 2705 2794
Corus A, 2006

8
Viswanathan Anand
 9/14 (64%) 2728 2791
Mexico (World Championship), 2007

9
Vladimir Kramnik
6.5/9 (72%) 2717 2788
Tal Memorial, Moscow, 2007

10
Veselin Topalov
 7/10 (70%) 2717 2786
Nanjing, 2008

And...:

Best Tournament Performances of All Time# Player Score/% opp. Perf. Event

1
Anatoly Karpov
11/13 (85%) 2729 2899
Linares, 1994

2
Garry Kasparov
12/14 (86%) 2692 2881
Tilburg, 1989

3
Emanuel Lasker
18/22 (82%) 2667 2878
London, 1899

4
Garry Kasparov
10.5/14 (75%) 2758 2877
Linares, 1999

5
Mikhail Tal
20/28 (71%) 2716 2869
Bled/Zagreb/Belgrade (Cand.), 1959

6
Alexander Alekhine
13/14 (93%) 2626 2865
San Remo, 1930

7
Garry Kasparov
10/13 (77%) 2737 2863
Linares, 1993

8
Alexander Alekhine
19.5/24 (81%) 2644 2859
Bled, 1931

9
Garry Kasparov
11.5/15 (77%) 2715 2856
Belfort (World Cup), 1988

10
Garry Kasparov
10/13 (77%) 2728 2855
Linares, 1992

11
Emanuel Lasker
11.5/16 (72%) 2738 2853
St. Petersburg, 1914

12
Garry Kasparov
 9/12 (75%) 2744 2851
Amsterdam (Optiebeurs), 1988

13
Garry Kasparov
9.5/11 (86%) 2682 2850
Belgrade (Investbank), 1989
 
Bobby Fischer
18.5/23 (80%) 2643 2850
Palma de Mallorca (Interzonal), 1970
 
Mikhail Botvinnik
14/20 (70%) 2729 2850
The Hague/Moscow (WCh), 1948
 
Magnus Carlsen
 8/10 (80%) 2733 2850
Nanjing, 2009

17
Siegbert Tarrasch
29/39 (74%) 2650 2846
Vienna, 1898

18
Garry Kasparov
8.5/11 (77%) 2733 2845
Linares, 1997

19
Johannes Zukertort
22.5/29 (78%) 2641 2844
London, 1883

20
Garry Kasparov
11/14 (79%) 2691 2840
Niksic, 1983

21
Vassily Ivanchuk
 9.5/13 (73%) 2732 2837
Linares, 1991
 
Géza Maróczy
16.5/22 (75%) 2671 2837
Ostend, 1905
 
Paul Keres
18.5/28 (66%) 2719 2836
Bled/Zagreb/Belgrade (Cand.), 1959
 
Garry Kasparov
10/13 (77%) 2705 2836
Wijk aan Zee (Hoogovens), 1999

25
Vassily Ivanchuk
 8/10 (80%) 2715 2835
Mtel Masters, Sofia, 2008

More clearly formatted on the Chessbase site.

bump...

just because it's interesting

Woot! Go Carlsen! He will be world chess champion! He will break Kasparov's FIDE rating record! Keep on winning man!!!

I'm not sure what tournaments were included in the compilation, but what about Sofia Polgar's famous 2900+ performance (at age 14) with 8.5/9 at  il Torneo Magistrale di Roma. in 1989?

It's almost certain that the way Sonas's system works (not ELO) that her performance doesn't qualify. Likely the strength of the opposition--same reason Fischer's 11-0 US Championship sweep doesn't rank high.

Just my surmise.

edit: He sees her performance as 2735

http://db.chessmetrics.com/CM2/SingleEvent.asp?Params=199510SSSSS3S102714000000121100877800024610100

Thanks. I figured there was some delimiter in the picking and choosing of tournaments.  The 2900+ figure comes from Polgar's own site

No doubt it is an accurate ELO number. I've seen it referenced many times over the years.

For that matter, Fischer's 6-0 sweep of Taimanov and Larsen combined has been quoted above 3000 as an ELO number.

I get the impression that scores without losses are tricky to gauge in the ELO system. Maybe a statistican can address this here in the forums sometime.

I think you're right, goldendog - it seems statistically very tricky to gauge 100% scores. I wonder if this is because it's hard to pin down exact numbers to the performance. With any score above 0% and below 100%, there's context; a definite fixed point where the performance could be improved or worsened; whereas with 100% it could be 2900, 3100, 3300... there's no limiter.